NVIDIA GeForce GTX 870M vs NVIDIA GeForce GTX 960A Comparison

The Verdict

The database comparison between the NVIDIA GeForce GTX 960A and the NVIDIA GeForce GTX 870M is defined by a narrow but consistent margin in the only shared benchmark. The GTX 870M wins the Geekbench OpenCL test with a score of 12630, while the GTX 960A records 11998, placing the 870M 5% ahead. The GTX 960A holds a higher percentile rank at 51, versus 48 for the 870M, but that ranking reflects its position against all GPUs in the database, not a direct performance advantage in this matchup. For a user choosing between these two MXM modules, the benchmark data points to the GTX 870M as the faster option in compute workloads, provided the host system can accommodate its higher power draw. The GTX 960A remains viable for those prioritizing a lower power envelope, as its 75 W TDP is 25 W below the 870M’s 100 W rating. Neither card is current, both are marked end-of-life, and neither has a recorded launch MSRP in the database.

Architecture Differences

The two NVIDIA parts come from different architectural generations and are built on different silicon. The GTX 960A uses the GM107 chip on the Maxwell architecture, fabricated by TSMC on a 28 nm process. The GTX 870M uses the GK104 chip on the older Kepler architecture, also from TSMC at 28 nm. This process parity contrasts sharply with the transistor counts: the GK104 packs 3,540 million transistors on a 294 mm² die, while the GM107 contains 1,870 million transistors on a 148 mm² die. Transistor density is nearly identical, with 12.6 million per mm² for the 960A and 12.0 million per mm² for the 870M, so the 870M’s larger die is simply a larger implementation of a similar density.

The 870M’s compute resources are substantially higher. It carries 1,344 shading units, 112 texture mapping units, and 24 ROPs. The 960A has 640 shading units, 40 TMUs, and 16 ROPs. This 2.1x difference in shading units and 2.8x difference in TMUs explains why the 870M posts higher theoretical rates: its pixel rate is 27.08 GPixel/s versus 18.82 GPixel/s for the 960A, and its texture rate is 108.3 GTexel/s versus 47.04 GTexel/s. Floating point throughput also favors the 870M at 2.599 TFLOPS, which is 73% higher than the 960A’s 1.505 TFLOPS.

Clock speeds tell the opposite story. The 960A has a base clock of 1097 MHz and a boost clock of 1176 MHz, while the 870M runs at 941 MHz base and 967 MHz boost. The 960A’s Maxwell architecture also supports Vulkan 1.4, whereas the 870M supports Vulkan 1.2.175; both are limited to DirectX 12 (11_0) and OpenGL 4.6. Memory configurations differ as well: the 960A has 2 GB of GDDR5 on a 128 bit bus with 80.19 GB/s bandwidth, while the 870M has 3 GB of GDDR5 on a 192 bit bus with 120.0 GB/s bandwidth, a 50% bandwidth advantage. Both use MXM-B (3.0) interfaces, have no power connectors, and list display outputs as portable device dependent. The 870M’s predecessor is the GeForce 700M series and its successor is the GeForce 900M series; the 960A’s predecessor is the GeForce 800A series, with no successor listed.

Head-to-Head Benchmarks

The Geekbench OpenCL result is the only directly comparable benchmark in the database, and it favors the GTX 870M by 5%. The 870M scores 12630, while the 960A scores 11998. That 632 point gap is modest in absolute terms, but it aligns with the 870M’s larger compute core count and higher memory bandwidth. The 960A counters with a higher percentile standing at 51 versus 48, but percentiles are relative to all GPUs, not just this pair.

Looking at nearest rivals in the database provides context for each card’s standing. The GTX 960A’s average benchmark score of 11998 places it just 0.3% above the NVIDIA GeForce GTX 1080 at 11960, and 1.3% above the AMD Radeon RX 6500 XT at 11842. It also sits 2.7% ahead of the NVIDIA GeForce GTX 1660 at 11680 and 3.2% ahead of the AMD Radeon RX 7800 XT at 11627. These deltas are all small, suggesting the 960A sits in a tightly packed performance cluster. The GTX 870M’s average score is 9959, which is 0.5% below the AMD Radeon Pro 5300M at 10013, 0.8% below the NVIDIA Quadro K5100M at 10043, and 1.1% below the AMD Radeon R9 M375 at 10070. It is 1.1% above the NVIDIA Quadro 6000 at 9846. The 870M’s rivals are all within a 1.1% band, indicating that its compute performance is highly competitive with those parts.

The 870M also has a Metal benchmark score of 7288, but the 960A has no Metal result in the database, so no cross-card comparison is possible on that API. The single OpenCL head-to-head remains the decisive data point, and it gives the 870M the only win in this comparison.

FAQ

Q: Which GPU is faster in compute workloads?

A: The GTX 870M, based on the Geekbench OpenCL result. It scores 12630 versus the GTX 960A’s 11998, a 5% lead.

Q: Does the GTX 960A have any advantage in performance rankings?

A: Yes. The GTX 960A sits at the 51st percentile among all GPUs in the database, while the GTX 870M sits at the 48th percentile. This reflects their overall standings, not their direct comparison.

Q: How do their memory subsystems compare?

A: The GTX 870M has 3 GB of GDDR5 on a 192 bit bus with 120.0 GB/s bandwidth. The GTX 960A has 2 GB of GDDR5 on a 128 bit bus with 80.19 GB/s bandwidth. The 870M offers 50% more bandwidth.

Q: Which card uses less power?

A: The GTX 960A, with a TDP of 75 W. The GTX 870M is rated at 100 W, making the 960A the lower-power option by 25 W.

Q: Do both cards support the same API level?

A: Both support DirectX 12 (11_0) and OpenGL 4.6. They differ on Vulkan: the GTX 960A supports Vulkan 1.4, while the GTX 870M supports Vulkan 1.2.175.

Q: How do their nearest rivals compare for each card?

A: The GTX 960A is within 3.2% of its closest rivals, with the GTX 1080 at 0.3% above and the RX 7800 XT at 3.2% below. The GTX 870M is within 1.1% of its rivals, with the Radeon Pro 5300M at 0.5% above and the Quadro 6000 at 1.1% below.

Where Each One Wins

The GTX 870M wins the only direct benchmark comparison. Its OpenCL score of 12630 exceeds the 960A’s 11998 by 5%, and its theoretical compute metrics reinforce that lead. With 2.599 TFLOPS of FP32 performance, 108.3 GTexel/s texture rate, and 27.08 GPixel/s pixel rate, the 870M is the stronger part for raw number crunching. Its 120.0 GB/s memory bandwidth and 3 GB framebuffer also give it an edge in memory-intensive workloads, particularly those that exceed 2 GB of video memory usage. The 870M’s 1344 shading units and 112 TMUs provide a wide execution width that should handle parallel compute tasks more efficiently than the 960A’s 640 shading units and 40 TMUs.

The GTX 960A’s wins are architectural and efficiency-based rather than performance-based. Its 75 W TDP is 25 W lower than the 870M’s, which matters in thin or thermally constrained portables using MXM modules. The 960A also runs at higher clocks, with a 1097 MHz base and 1176 MHz boost versus 941 MHz and 967 MHz for the 870M, which can help in lightly threaded or latency-sensitive scenarios. Its Vulkan 1.4 support is a newer API revision than the 870M’s Vulkan 1.2.175, and its higher percentile rank of 51 versus 48 indicates better positioning in the overall GPU landscape. The 960A’s smaller die, 148 mm² versus 294 mm², and lower transistor count, 1,870 million versus 3,540 million, also point to lower manufacturing complexity, though that is a production consideration rather than a user-facing feature.

For a specific use case, the data suggests choosing the GTX 870M when compute performance is the priority and the system can handle its 100 W TDP. The 960A is the pick when power draw is the limiting factor, as its 75 W TDP and higher clocks offer a more efficient profile, though with 5% less OpenCL performance. In the absence of any other head-to-head benchmarks, the OpenCL result is the sole basis for a performance verdict, and it favors the older Kepler part by a clear margin.